Sponsored byFusecodeEnterprise coding API for Claude Code, Codex, and model workflows.
LogoLMSpeed
  • Free
  • Models
  • Providers
  • Leaderboard
  • Docs
LogoLMSpeed
  1. Home
  2. Leaderboard
  3. Model Speed
LogoLMSpeed

The best API speed test tool

GitHubGitHubTwitterX (Twitter)Email
Product
  • Features
  • Pricing
  • FAQ
Leaderboard
  • Overview
  • Speed Ranking
  • Latency Ranking
  • Health Ranking
  • Model Pricing
  • Model Speed
  • Reasoning
  • Coding
Models
  • All Models
  • GPT
  • Claude
  • Gemini
  • DeepSeek
  • Llama
  • Qwen
Free Models
  • All Free Models
  • Free GPT
  • Free Claude
  • Free Gemini
  • Free DeepSeek
  • Free Llama
  • Free Qwen
Tools
  • Speed Test
  • Provider Audit
Company
  • About
Resources
  • Provider Directory
  • Documentation
  • Public API
  • Botab
  • VidBee
Legal
  • Cookie Policy
  • Privacy Policy
  • Terms of Service
© 2026 LMSpeed All Rights Reserved.Made by Nexmoe with ❤️
AgentsCodingReasoningKnowledgeMathMultilingualMultimodalInstruction following

LMSpeed Model Speed Leaderboard

Multi-dimensional rankings based on model speed tests, provider health checks, and standard model benchmarks. Compare providers, endpoints, models, and reliability at a glance.

Compares AI model responsiveness across Artificial Analysis output speed, time to first token, and time to first answer token in one table.

Updated: Jul 28, 2026, 04:45 AM·Artificial Analysis standard benchmarks

Fairness Notice
These model leaderboards use imported Artificial Analysis benchmark records and are deduplicated to one canonical LMSpeed model per rank.
RankModelOutput SpeedTime to First TokenTime to First Answer TokenUpdated
1
Mercury 2mercury-2
1074.7 tok/s#1
4.15 s#53
4.15 s#25
Jul 28, 2026, 04:45 AM
2
Stepfun
Step 3.7 Flashstep-3-7-flash
419.3 tok/s#2
0.72 s#20
5.50 s#27
Jul 28, 2026, 04:45 AM
3
Gemini
Gemini 3.1 Flash Lite Previewgemini-3-1-flash-lite-preview
334.2 tok/s#3
5.45 s#54
5.45 s#26
Jul 28, 2026, 04:45 AM
4
Nvidia
Nemotron 3 Nano Omninemotron-3-nano-omni-30b-a3b-reasoning
320.4 tok/s#4
0.52 s#10
6.76 s#29
Jul 28, 2026, 04:45 AM
5
OpenAI
gpt-oss-120bgpt-oss-120b
310.3 tok/s#5
0.55 s#11
6.99 s#30
Jul 28, 2026, 04:45 AM
6
Stepfun
Step 3.5 Flashstep-3-5-flash
292.6 tok/s#6
0.81 s#21
7.64 s#31
Jul 28, 2026, 04:45 AM
7
Gemini
Gemini 3.5 Flashgemini-3-5-flash
265.6 tok/s#7
12.68 s#58
12.68 s#37
Jul 28, 2026, 04:45 AM
8
Qwen
Qwen3.5 Omni Flashqwen3-5-omni-flash
261.4 tok/s#8
0.97 s#29
0.97 s#14
Jul 28, 2026, 04:45 AM
9
OpenAI
GPT-5.6 Lunagpt-5-6-luna
220.4 tok/s#9
84.29 s#64
84.29 s#62
Jul 28, 2026, 04:45 AM
10
Nvidia
Nemotron 3 Ultranemotron-3-ultra-550b-a55b
219.2 tok/s#10
0.86 s#24
11.24 s#36
Jul 28, 2026, 04:45 AM
11
ChatGLM
GLM-5.2glm-5-2
219.2 tok/s#11
0.81 s#22
9.94 s#33
Jul 28, 2026, 04:45 AM
12
OpenAI
gpt-oss-20bgpt-oss-20b
209.3 tok/s#12
0.47 s#7
10.02 s#34
Jul 28, 2026, 04:45 AM
13
Qwen
Qwen3 Next 80B A3B Instructqwen3-next-80b-a3b-instruct
205.7 tok/s#13
1.00 s#31
1.00 s#16
Jul 28, 2026, 04:45 AM
14
Qwen
Qwen3.7 Maxqwen3-7-max
203.1 tok/s#14
1.74 s#46
13.60 s#39
Jul 28, 2026, 04:45 AM
15
Qwen
Qwen3.5-35B-A3Bqwen3-5-35b-a3b
172.9 tok/s#15
1.02 s#32
1.02 s#17
Jul 28, 2026, 04:45 AM
16
Hunyuan
Hy3 previewhy3-preview
165.5 tok/s#16
2.01 s#49
14.10 s#40
Jul 28, 2026, 04:45 AM
17
OpenAI
O3o3
164.6 tok/s#17
5.92 s#55
5.92 s#28
Jul 28, 2026, 04:45 AM
18
Trinity Large Thinkingtrinity-large-thinking
162.5 tok/s#18
0.48 s#9
12.79 s#38
Jul 28, 2026, 04:45 AM
19
Qwen
Qwen3.5-122B-A10Bqwen3-5-122b-a10b
156.3 tok/s#19
1.05 s#33
1.05 s#18
Jul 28, 2026, 04:45 AM
20
Gemini
Gemini 2.5 Progemini-2-5-pro
153.2 tok/s#20
19.85 s#59
19.85 s#43
Jul 28, 2026, 04:45 AM
21
OpenAI
GPT-5.3 Codexgpt-5-3-codex
148.4 tok/s#21
60.34 s#62
60.34 s#60
Jul 28, 2026, 04:45 AM
22
Gemini
Gemini 3.1 Pro Previewgemini-3-1-pro-preview
146.5 tok/s#22
26.14 s#60
26.14 s#47
Jul 28, 2026, 04:45 AM
23
Qwen
Qwen3 Coder Nextqwen3-coder-next
143.8 tok/s#23
0.84 s#23
0.84 s#12
Jul 28, 2026, 04:45 AM
24
Nex n2 Pronex-n2-pro
142.1 tok/s#24
0.88 s#26
14.96 s#41
Jul 28, 2026, 04:45 AM
25
Ling 2.6 Flashling-2-6-flash
142.0 tok/s#25
0.61 s#16
0.61 s#9
Jul 28, 2026, 04:45 AM
26
Ring 2 6 1tring-2-6-1t
135.1 tok/s#26
1.80 s#47
16.60 s#42
Jul 28, 2026, 04:45 AM
27
OpenAI
GPT-5.6 Terragpt-5-6-terra
134.9 tok/s#27
0.67 s#18
0.67 s#10
Jul 28, 2026, 04:45 AM
28
Claude
Claude Haiku 4.5claude-haiku-4-5
129.1 tok/s#28
0.61 s#15
0.61 s#8
Jul 28, 2026, 04:45 AM
29
DeepSeek
DeepSeek V4 Flashdeepseek-v4-flash
125.4 tok/s#29
0.95 s#28
45.72 s#56
Jul 28, 2026, 04:45 AM
30
MetaAI
Llama 4 Maverickllama-4-maverick
121.7 tok/s#30
0.60 s#14
0.60 s#7
Jul 28, 2026, 04:45 AM
31
Granite 4.1 8Bgranite-4-1-8b
118.2 tok/s#31
0.44 s#5
0.44 s#3
Jul 28, 2026, 04:45 AM
32
Grok
Grok 4.3grok-4-3
114.5 tok/s#32
0.87 s#25
0.87 s#13
Jul 28, 2026, 04:45 AM
33
Qwen
Qwen3.6 35B A3Bqwen3-6-35b-a3b
107.9 tok/s#33
1.41 s#40
51.39 s#57
Jul 28, 2026, 04:45 AM
34
Mistral
Mistral Medium 3.5mistral-medium-3-5
101.4 tok/s#34
0.59 s#13
20.32 s#44
Jul 28, 2026, 04:45 AM
35
Kat Coder Pro V2kat-coder-pro-v2
100.9 tok/s#35
1.45 s#42
1.45 s#22
Jul 28, 2026, 04:45 AM
36
MetaAI
Llama 4 Scoutllama-4-scout
97.7 tok/s#36
0.56 s#12
0.56 s#6
Jul 28, 2026, 04:45 AM
37
Reka Flash 3reka-flash-3
93.9 tok/s#37
2.42 s#51
23.72 s#45
Jul 28, 2026, 04:45 AM
38
Claude
Claude Sonnet 5claude-sonnet-5
85.7 tok/s#38
109.53 s#65
109.53 s#65
Jul 28, 2026, 04:45 AM
39
Minimax
MiniMax M3minimax-m3
85.5 tok/s#39
1.19 s#34
24.59 s#46
Jul 28, 2026, 04:45 AM
40
OpenAI
GPT-5.6 Solgpt-5-6-sol
81.8 tok/s#40
8.65 s#56
8.65 s#32
Jul 28, 2026, 04:45 AM
41
DeepSeek
DeepSeek V4 Prodeepseek-v4-pro
75.2 tok/s#41
0.95 s#27
27.44 s#48
Jul 28, 2026, 04:45 AM
42
MiMo-V2.5-Promimo-v2-5-pro
72.5 tok/s#42
2.01 s#49
29.60 s#49
Jul 28, 2026, 04:45 AM
43
MiMo-V2.5mimo-v2-5
72.1 tok/s#43
3.04 s#52
30.79 s#51
Jul 28, 2026, 04:45 AM
44
Qwen
Qwen3.5 397B A17Bqwen3-5-397b-a17b
71.8 tok/s#44
1.45 s#43
1.45 s#23
Jul 28, 2026, 04:45 AM
45
Claude
Claude Fable 5claude-fable-5
71.3 tok/s#45
80.06 s#63
80.06 s#61
Jul 28, 2026, 04:45 AM
46
Ministral 3ministral-3
70.0 tok/s#46
0.46 s#6
0.46 s#4
Jul 28, 2026, 04:45 AM
47
Qwen
Qwen3.5-9Bqwen3-5-9b
67.2 tok/s#47
0.65 s#17
30.42 s#50
Jul 28, 2026, 04:45 AM
48
Claude
Claude Opus 4.8claude-opus-4-8
65.0 tok/s#48
37.44 s#61
37.44 s#53
Jul 28, 2026, 04:45 AM
49
Mistral
Mistral Large 3mistral-large-3
62.9 tok/s#49
0.69 s#19
0.69 s#11
Jul 28, 2026, 04:45 AM
50
Grok
Grok 4.5grok-4-5
61.2 tok/s#50
10.85 s#57
10.85 s#35
Jul 28, 2026, 04:45 AM
51
Qwen
Qwen3.6 27Bqwen3-6-27b
61.0 tok/s#51
1.41 s#40
94.52 s#63
Jul 28, 2026, 04:45 AM
52
Hunyuan
Hy3hy3
59.5 tok/s#52
1.34 s#39
34.96 s#52
Jul 28, 2026, 04:45 AM
53
Claude
Claude Sonnet 4.6claude-sonnet-4-6
59.1 tok/s#53
1.26 s#37
1.26 s#20
Jul 28, 2026, 04:45 AM
54
Cohere
Command Acommand-a
59.0 tok/s#54
0.48 s#8
0.48 s#5
Jul 28, 2026, 04:45 AM
55
Claude
Claude Opus 4.7claude-opus-4-7
56.8 tok/s#55
1.72 s#45
1.72 s#24
Jul 28, 2026, 04:45 AM
56
Qwen
Qwen3.5 Omni Plusqwen3-5-omni-plus
54.9 tok/s#56
1.24 s#36
1.24 s#19
Jul 28, 2026, 04:45 AM
57
Qwen
Qwen3.6 Plusqwen3-6-plus
52.5 tok/s#57
1.65 s#44
107.29 s#64
Jul 28, 2026, 04:45 AM
58
Qwen
Qwen3.7 Plusqwen3-7-plus
52.2 tok/s#58
2.01 s#48
40.32 s#54
Jul 28, 2026, 04:45 AM
59
Azure
Phi 4 Mini Instructphi-4-mini-instruct
46.7 tok/s#59
0.33 s#2
0.33 s#1
Jul 28, 2026, 04:45 AM
60
North Mini Codenorth-mini-code
46.4 tok/s#60
0.31 s#1
43.38 s#55
Jul 28, 2026, 04:45 AM
61
MoonshotAI
Kimi K2.7 Codekimi-k2-7-code
42.6 tok/s#61
1.21 s#35
53.55 s#58
Jul 28, 2026, 04:45 AM
62
MoonshotAI
Kimi K2.6kimi-k2-6
35.7 tok/s#62
1.29 s#38
1.29 s#21
Jul 28, 2026, 04:45 AM
63
Qwen
Qwen3.5qwen3-5
34.7 tok/s#63
0.39 s#4
58.08 s#59
Jul 28, 2026, 04:45 AM
64
MoonshotAI
Kimi K3kimi-k3
32.9 tok/s#64
125.71 s#66
186.48 s#66
Jul 28, 2026, 04:45 AM
65
Devstral 2devstral-2
31.5 tok/s#65
0.97 s#30
0.97 s#15
Jul 28, 2026, 04:45 AM
66
Azure
Phi 4 Multimodal Instructphi-4-multimodal-instruct
18.0 tok/s#66
0.35 s#3
0.35 s#2
Jul 28, 2026, 04:45 AM
Throughputt/s
First Token Latencys
HealthAvg probe latency
Model PerformanceAA benchmark matrix
Model PricingAA pricing matrix
Model
Provider